DE102019101045B4 - Device and method for regulating a vehicle - Google Patents

Abstract

Method for regulating a vehicle, characterized in that a model predictive controller specifies at least one manipulated variable (u) for at least one of the plurality of controlled systems for influencing the driving behavior of a vehicle, the at least one manipulated variable (u) depending on a time-discrete dynamic model of the vehicle to be controlled is determined (204), with a trajectory of future states of the model up to a time horizon being determined (204) for a current state of the model, the trajectory depending on a change in the at least one position variable (Δu (k)) is determined, and the change in the at least one manipulated variable (Δu (k)) for determining the trajectory is approximated by means of at least one Laguerre function, characterized in that the trajectory is dependent on a quality measure (J ) is determined, wherein the quality measure (J) depends on a control deviation (e) and a control effort, and that the quality measure (J) is defined as a function of at least one limit value which characterizes a limitation of an actuating range of an actuator or a limitation for the value of a manipulated variable (u) for at least one of the plurality of controlled systems.

Description

The invention is based on a method for regulating a vehicle according to the preamble of claim 1. Furthermore, the invention relates to a device for regulating a vehicle, in particular a vehicle with an electric drive, according to the preamble of claim 6.

The publication "Adel, EM, Oulandsine, M., & Radouane, L .: Predictive steering control using Laguerre series representation. Proceedings of 2003 IEEE Conference on Control Applications, 203. CCA 2003., 1, 439-445 vol. 1. ‟ already discloses a method for controlling a vehicle according to the preamble of patent claim 1.

• - das Bereitstellen eines mathematischen Modells der Fahrzeugdynamik, das eine Zustandsvariable, eine Lenksteuerungsvariable und einen zukünftigen Fahrbahnstörungsfaktor beinhaltet, der eine Krümmung, einen Hang und/oder eine Steigung der Fahrbahn definiert;
• - das Bestimmen eines Lenksteuerungsziels unter Verwendung der Lenksteuerungsvariable, die die Differenz zwischen einem aktuellen Fahrzeugweg und einem gewünschten Fahrzeugweg verringert;
• - das Bestimmen eines optimalen Lenksteuersignals unter Verwendung des mathematischen Modells, das das Lenksteuerungsziel bereitstellt und das einen Rückkopplungsabschnitt und einen Vorwärtskopplungsabschnitt beinhaltet, wobei der Vorwärtskopplungsabschnitt den Straßenstörungsfaktor beinhaltet; und
• - das Bereitstellen des Steuersignals an eine Lenkungssteuerung.

DE 10 2017 201 569 A1 discloses a method of providing lateral steering control for an autonomously powered or semi-autonomously powered vehicle traveling along a roadway, the method comprising:

• the provision of a mathematical model of the vehicle dynamics that includes a state variable, a steering control variable and a future road surface disturbance factor which defines a curvature, a slope and / or a gradient of the road;
• - determining a steering control target using the steering control variable that reduces the difference between an actual vehicle path and a desired vehicle path;
• - determining an optimal steering control signal using the mathematical model that provides the steering control target and that includes a feedback section and a feedforward section, the feedforward section including the road disturbance factor; and
• the provision of the control signal to a steering controller.

The increasing number of control systems and assistance systems in vehicles of this type with electric drive makes the peaceful coexistence of these systems increasingly complex. The effort for the application of these systems increases. Vehicles with electric drive, in particular vehicles with wheel hub motors, also offer completely new degrees of freedom, for example for torque vectoring. The effectiveness of the control systems and assistance systems can be increased by coordinated control by a central controller. Closed consideration of the limits of the control systems is essential for effective, centralized control.

Integrated driving dynamics controllers are used to enable central control.

US 2018/0 079 272 A1 discloses a method in which a road profile of a road at a particular location is made available in a control system of a vehicle. The method provides that an active suspension system of the vehicle is adapted when the vehicle approaches the specific location.

DE 10 2007 019 053 A1 discloses a method of controlling a vehicle having a suspension, the method including adjusting a suspension element during longitudinal vibrations of the vehicle to vary a normal force between the vehicle and the surface. In particular, a procedure is described in which the active suspension uses "online learning" as well as predictive control with the aid of a model.

DE 10 2014 200 031 A1 describes a method for adaptively and actively controlling a suspension system of a vehicle with road preview, determining a degree of a road abnormality in front of the vehicle and controlling the suspension in response to the type and severity of the abnormality. In particular, a model-predictive control is provided for this purpose.

It is desirable to improve such systems further.

This is achieved by the device and the method according to the independent claims.

The method provides that a model-predictive controller specifies at least one manipulated variable for at least one of the plurality of controlled systems for influencing the driving behavior of a vehicle, the at least one manipulated variable being determined as a function of a time-discrete dynamic model of the vehicle to be controlled, wherein a trajectory of future states of the model up to a time horizon is determined for a current state of the model, the trajectory being determined as a function of a change in the at least one manipulated variable, and the change in the at least one manipulated variable for determining the trajectory by means of at least one Laguerre function is approximated. It is also provided that the trajectory is determined as a function of a quality measure, the quality measure depending on a control deviation and a control effort, and that the quality measure is defined as a function of at least one limit value, which is a limitation of a setting range of an actuator or a limitation for the value a manipulated variable for at least one of the large number of controlled systems characterized. Based on an internal vehicle model, the future vehicle states are predicted and the control system interventions are calculated. The controller is coordinated from the model knowledge, ie via the effect of the control systems of the model predictive controller. The computing time can be reduced using Laguerre functions. This means that the model predictive controller can also be used in a control unit for a vehicle.

The quality measure represents a defined criterion with which optimal control system interventions are determined. This enables a minimization of the control system interventions, for example in order to avoid unnecessary interventions, and a minimization of the deviation of the actual trajectory from a target trajectory.

Since the quality measure is defined as a function of at least one limit value, which characterizes a limitation of an adjustment range of an actuator or a limitation for the value of a manipulated variable for at least one of the large number of controlled systems, limits that are set by the actuators of the vehicle or the physical conditions 5 for a regulation are given.

The model-predictive controller preferably comprises a time-discrete state prediction which is defined by a linearized time-invariant state model and Laguerre coefficients. In contrast to conventional model predictive controllers, the Laguerre coefficients are optimized. This represents a 0 prediction that can be calculated particularly quickly.

The model predictive controller is preferably defined as a function of the change in the at least one manipulated variable for each time step up to a time horizon.

Different manipulated variables preferably influence different vehicle components from the drive, rear axle steering, control unit and / or actuators 5 for torque vectoring, all-wheel drive system, chassis actuator, in particular in autonomous driving mode.

The at least one manipulated variable is preferably determined as a function of a reference variable.

The device comprises a microprocessor and a memory with instructions, which, when executed by the microprocessor, execute the method.

The device preferably comprises a first interface for at least one controlled variable and a second interface for at least one manipulated variable, the first interface being designed to receive information about controlled variables from different controlled systems, the second interface being designed to output manipulated variables for different controlled systems.

• 1 schematisch eine Vorrichtung zur zentralen Regelung eines Fahrzeugs,
• 2 schematisch ein Flussdiagram mit Schritten eines Verfahrens zur Regelung des Fahrzeugs.

Further advantageous refinements emerge from the following description and the drawing. In the drawing shows

• 1 schematically a device for the central control of a vehicle,
• 2 schematically a flow diagram with steps of a method for regulating the vehicle.

1 shows schematically an apparatus 100 for the central control of a vehicle. In the example, the vehicle is an all-wheel drive vehicle with rear-axle steering that is driven by an electric motor. The device 100 includes a microprocessor 102 and a memory 104 .

The device 100 includes a first interface 106 for at least one controlled variable y and a second interface 108 for at least one manipulated variable u, the first interface 106 is designed to receive information about controlled variables y from different controlled systems, the second interface 108 is designed to output manipulated variables u for different controlled systems.

In the example, the controlled systems include different actuators, the adjustment ranges of which are restricted by limitations. A limitation can also be provided which characterizes the value of a manipulated variable for at least one of the plurality of controlled systems. This takes the physical conditions into account.

The information about controlled variables y from different controlled systems is recorded via sensors or determined by control units that work with the device 100 communicate via a communication network of the vehicle. The manipulated variables u for different controlled systems are output to the actuators or control devices for controlling the different controlled systems via the communication network. In the example, the communication network is a controller area network, CAN.

In the example, a desired yaw rate, longitudinal speed and lateral speed are to be set. In the example, a corresponding reference variable is determined depending on a single-track model for the vehicle. For example, a steering angle δ _F for the single-track model is dependent on a target steering angle as a control variable that is supplied by a steering angle sensor, and an actual speed v of the vehicle, which is provided by a control unit of the vehicle supplies, used to determine a desired yaw rate.

mit
c_sRSchräglaufsteifigkeit vorne rechts,
c_sLSchräglaufsteifigkeit vorne links.To do this, the characteristic speed of the vehicle is determined: $${\nu }_{cH}=\left(l_{\mathrm{F.}}+l_{\mathrm{R.}}\right)\sqrt{\frac{c_{s\mathrm{F.}{c}_{s\mathrm{R.}}}}{\mathrm{M.}\left(l_{\mathrm{R.}}{c}_{s\mathrm{R.}}-l_{\mathrm{F.}}{c}_{s\mathrm{F.}}\right)}}$$

With
c _{sR slip rigidity} front right,
c _{sL Skew} stiffness front left.

mit δ_FLenkwinkel des Einspurmodells.Depending on this, a target yaw rate is determined: $${\psi }_{des\mathrm{,1}}=\frac{\nu {\delta }_{\mathrm{F.}}}{\left(l_{\mathrm{F.}}+l_{\mathrm{R.}}\right)\left(1+\frac{{\nu }^2}{{\nu }_{cH}^2}\right)}$$

with δ _F steering angle of the single-track model.

In order to avoid unrealistically high values, the target yaw rate is limited depending on the coefficient of friction µ, the acceleration due to gravity g and the speed of the vehicle v to: $${\psi }_{des}=min\left({\dot{\psi }}_{des\mathrm{,1}},\frac{\mu G}{\nu }\right)$$

A steady state with β̇̇̇̇̇ = 0 is assumed for a target float angle. The target float angle is therefore dependent on the yaw rate Ψ̇: $${\beta }_{des}=\frac{c_{s\mathrm{F.}}{\delta }_{\mathrm{F.}}-\left(\mathrm{M.}{\nu }^2+c_{s\mathrm{F.}}{l}_{\mathrm{F.}}-c_{s\mathrm{R.}}{l}_r\right)\frac{\dot{\psi }}{\nu }}{c_{s\mathrm{F.}}+c_{s\mathrm{R.}}}$$

The target float angle is limited to 5 °, for example.

The memory 104 includes instructions when executed by the microprocessor 102 the procedure described below is followed.

$$y\left(k\right)=\left[0I\right]x^{aug}\left(k\right)$$

mit $$x^{aug}=\left[\Delta x_m^T{e}^T\right]$$

In the method, a model predictive controller is used which comprises the following discrete time dynamic model of the vehicle to be controlled: $$x^{erw}\left(k+1\right)=\left[\begin{array}{cc}{\mathrm{A.}}_m& 0^T\\ {\mathrm{C.}}_m{\mathrm{A.}}_m& \mathrm{I.}\end{array}\right]x^{auG}\left(k\right)+\left[\begin{array}{c}{\mathrm{B.}}_m\\ {\mathrm{C.}}_m{\mathrm{B.}}_m\end{array}\right]\Delta u\left(k\right)$$

$$y\left(k\right)=\left[0\mathrm{I.}\right]x^{auG}\left(k\right)$$

With $$x^{auG}=\left[\Delta x_m^T{e}^T\right]$$

wobei q Regelgrößen y und q Führungsgrößen y^des vorgesehen sind.In the procedure, the following system deviation is used in the example: $$e=\left[y_1-y_1^{des}{y}_2-y_2^{des}\cdots y_q-y_q^{des}\right]$$

where q controlled variables y and q reference variables y ^{des are} provided.

$$B_m=\frac{\partial f_d\left(x,u\right)}{\partial x}|{}_{x=x_m,u=u_m}$$

$$C_m=\frac{\partial f_d\left(x,u\right)}{\partial x}|{}_{x=x_m,u=u_m}$$

wobei die Übertragungsfunktionen f_d und g_d aus $$x\left(k+1\right)=f_d\left(x\left(k\right),u\left(k\right)\right)$$

$$y\left(k\right)=g_d\left(x\left(k\right),u\left(k\right)\right)$$

verwendet werden.The matrices of the model are defined as follows: $${\mathrm{A.}}_m=\frac{\partial f_d\left(x,u\right)}{\partial x}|{}_{x=x_m,u=u_m}$$

$${\mathrm{B.}}_m=\frac{\partial f_d\left(x,u\right)}{\partial x}|{}_{x=x_m,u=u_m}$$

$${\mathrm{C.}}_m=\frac{\partial f_d\left(x,u\right)}{\partial x}|{}_{x=x_m,u=u_m}$$

where the transfer functions f _d and g _d from $$x\left(k+1\right)=f_d\left(x\left(k\right),u\left(k\right)\right)$$

$$y\left(k\right)=G_d\left(x\left(k\right),u\left(k\right)\right)$$

be used.

mit den Laguerre Koeffizienten η_i und mit $$L_i{\left(k\right)}^T=\left[l_1^i\left(k\right)l_2^i\left(k\right)\dots l_3^i\left(k\right)\right].$$

wobei $$l_m\left(k\right)=Z^{-1}\left\{{\Gamma }_m\left(z\right)\right\}$$

mit $${\Gamma }_m\left(z\right)=\frac{\sqrt{1-a^2}}{1-a{z}^{-1}}{\left(\frac{z^{-1}-a}{1-a{z}^{-1}}\right)}^{N-1}$$

The time-discrete dynamic model of the vehicle to be controlled includes a vector of the change in the manipulated variables Δu _i (k), which is determined in the example as $$\Delta u_i\left(k\right)={\mathrm{L.}}_i{\left(k\right)}^T{\eta }_i$$

with the Laguerre coefficients η _i and with $${\mathrm{L.}}_i{\left(k\right)}^T=\left[l_1^i\left(k\right)l_2^i\left(k\right)...l_3^i\left(k\right)\right].$$

in which $$l_m\left(k\right)=Z^{-1}\left\{{\Gamma }_m\left(z\right)\right\}$$

With $${\Gamma }_m\left(z\right)=\frac{\sqrt{1-a^2}}{1-a{z}^{-1}}{\left(\frac{z^{-1}-a}{1-a{z}^{-1}}\right)}^{N-1}$$

The number of Laguerre coefficients η _i can be adapted according to the number i of manipulated variables. For the pole a of the discrete Laguerre network with order m obtained by inverse z-transformation of a continuous Laguerre network, in the example: $$0\le a\le 1$$

The pole a is a selectable scaling factor that influences the quality of the approximation.

und der Vektor $$u={\left[{\sigma }_{x_{FL}}{\sigma }_{x_{FR}}{\sigma }_{x_{RL}}{\sigma }_{x_{RR}}{\delta }_R\right]}^T$$

verwendet, mit
σ_{x FL} Längsschlupf des vorderen linken Rads,
σ_{x FR} Längsschlupf des vorderen rechten Rads,
σ_{x FL} Längsschlupf des hinteren linken Rads,
σ_{x RR} Längsschlupf des hinteren rechten Rads, und
δ_R Hinterachslenkwinkel.In the example, an optimal control trajectory is to be achieved for the vehicle depending on a system model of the vehicle. In the example, the state vector $$x={\left[{\nu }_x{\nu }_y\dot{\psi }\right]}^T$$

and the vector $$u={\left[{\sigma }_{x_{\mathrm{F.}\mathrm{L.}}}{\sigma }_{x_{\mathrm{F.}\mathrm{R.}}}{\sigma }_{x_{\mathrm{R.}\mathrm{L.}}}{\sigma }_{x_{\mathrm{R.}\mathrm{R.}}}{\delta }_{\mathrm{R.}}\right]}^T$$

used with
σ _{x FL} Longitudinal slip of the front left wheel,
σ _{x FR} Longitudinal slip of the front right wheel,
σ _{x FL} Longitudinal slip of the rear left wheel,
σ _{x RR} Rear right wheel longitudinal slip, and
δ _R rear axle steering angle.

In the example, the vector u is dimensioned for the control of the wheel slips and the rear axle control angle. For a large number of controlled systems, the vector u is dimensioned, for example, according to the number of manipulated variables available.

mit $$\varphi {\left(m\right)}^T={\displaystyle \sum _j^{m-1}{A}^{m-j-1}\left[B_1{L}_1{\left(j\right)}^T{B}_2{L}_2{\left(j\right)}^T\mathrm{...}{B}_3{L}_3{\left(j\right)}^T\right]}\eta$$

The following equation for calculating the trajectory of the state vector x (k + mlk) up to the horizon of the prediction N _P , ie for each step m <N _P after the point in time k, describes the time-discrete state prediction used in the controller using a linear one Time-invariant state model and with Laguerre coefficients with regard to the change in the manipulated variable Δu: $$x\left(k+m|k\right)={\mathrm{A.}}^{m}x\left(k\right)+\varphi {\left(m\right)}^T$$

With $$\varphi {\left(m\right)}^T={\displaystyle \sum _j^{m-1}{\mathrm{A.}}^{m-j-1}\left[{\mathrm{B.}}_1{\mathrm{L.}}_1{\left(j\right)}^T{\mathrm{B.}}_2{\mathrm{L.}}_2{\left(j\right)}^T\mathrm{...}{\mathrm{B.}}_3{\mathrm{L.}}_3{\left(j\right)}^T\right]}\eta$$

mit dem Zeithorizont N_P und $$\Omega ={\displaystyle \sum _{m=1}^{N_p}\varphi \left(m\right)Q\varphi {\left(m\right)}^T+R_L}$$

$$\psi ={\displaystyle \sum _{m=1}^{N_p}\varphi \left(m\right)Q{A}^m}$$

mit einer positiv semi-definiten Hesse-Matrix Ω, die das Auffinden eines globalen Minimums ermöglicht. Ohne Beschränkungen ist eine Optimierung analytisch lösbar. Werden Grenzen der Aktuatoren oder für die Stellgrößen berücksichtigt, wird ein beschränktes Optimierungsproblem wie folgt definiert: $$\underset{\eta }{{\displaystyle \text{min}}}J={\eta }^{T}Q\eta +2{\eta }^T\Psi x\left(k\right)$$

unter der Bedingung $$M\eta \le \gamma$$

In order to generate a satisfactory vehicle behavior, the following quality measure is used in the example, which is minimized in order to take into account both the control deviation and the control effort: $$J={\eta }^{T}Q\eta +2{\eta }^T\Psi x\left(k\right)+{\displaystyle \sum _{m=1}^{N_p}x{\left(k\right)}^T{\left({\mathrm{A.}}^T\right)}^{m}Q{\mathrm{A.}}^{m}x\left(k\right)}$$

with the time horizon N _P and $$\Omega ={\displaystyle \sum _{m=1}^{N_p}\varphi \left(m\right)Q\varphi {\left(m\right)}^T+{\mathrm{R.}}_{\mathrm{L.}}}$$

$$\psi ={\displaystyle \sum _{m=1}^{N_p}\varphi \left(m\right)Q{\mathrm{A.}}^m}$$

with a positive semi-definite Hessian matrix Ω, which enables a global minimum to be found. An optimization can be solved analytically without restrictions. If the limits of the actuators or the manipulated variables are taken into account, a limited optimization problem is defined as follows: $$\underset{\eta }{{\displaystyle \text{min}}}J={\eta }^{T}Q\eta +2{\eta }^T\Psi x\left(k\right)$$

under the condition $$\mathrm{M.}\eta \le \gamma$$

mit $$M_U=\left[\begin{array}{cccc}{\displaystyle {\sum }_{i=0}^{k-1}{L}_1{\left(i\right)}^T}& 0_2^T& & 0_m^T\\ 0_1^T& {\displaystyle {\sum }_{i=0}^{k-1}{L}_2{\left(i\right)}^T}& \cdots & 0_m^T\\ ⋮& ⋮& \ddots & ⋮\\ 0_1^T& 0_2^T& \cdots & {\displaystyle {\sum }_{i=0}^{k-1}{L}_m{\left(i\right)}^T}\end{array}\right]$$

und $$M_{\Delta U}=\left[\begin{array}{cccc}{L}_1{\left(m\right)}^T& 0_2^T& & 0_m^T\\ 0_1^T& L_2{\left(m\right)}^T& \cdots & 0_m^T\\ ⋮& ⋮& \ddots & ⋮\\ 0_1^T& 0_2^T& \cdots & L_m{\left(m\right)}^T\end{array}\right]$$

Where M and y describe the limit. In the example, a limit for a maximum desired wheel slip and rear axle steering angle is used. These restrictions limit the rate of change of the manipulated variables ΔU ^{(min / max)} and the absolute values U ^{(min / max)} as follows: $$\left[\begin{array}{c}{\mathrm{M.}}_{\Delta U}\\ -{\mathrm{M.}}_{\Delta U}\\ {\mathrm{M.}}_U\\ -{\mathrm{M.}}_U\end{array}\right]\eta \le \left[\begin{array}{c}\Delta U^{max}\\ \Delta U^{min}\\ U^{max}-U\left(k-1\right)\\ -\Delta U^{min}+U\left(k-1\right)\end{array}\right]$$

With $${\mathrm{M.}}_U=\left[\begin{array}{cccc}{\displaystyle {\sum }_{i=0}^{k-1}{\mathrm{L.}}_1{\left(i\right)}^T}& 0_2^T& & 0_m^T\\ 0_1^T& {\displaystyle {\sum }_{i=0}^{k-1}{\mathrm{L.}}_2{\left(i\right)}^T}& \cdots & 0_m^T\\ ⋮& ⋮& \ddots & ⋮\\ 0_1^T& 0_2^T& \cdots & {\displaystyle {\sum }_{i=0}^{k-1}{\mathrm{L.}}_m{\left(i\right)}^T}\end{array}\right]$$

and $${\mathrm{M.}}_{\Delta U}=\left[\begin{array}{cccc}{\mathrm{L.}}_1{\left(m\right)}^T& 0_2^T& & 0_m^T\\ 0_1^T& {\mathrm{L.}}_2{\left(m\right)}^T& \cdots & 0_m^T\\ ⋮& ⋮& \ddots & ⋮\\ 0_1^T& 0_2^T& \cdots & {\mathrm{L.}}_m{\left(m\right)}^T\end{array}\right]$$

The following based on the 2 The method described starts, for example, when the vehicle is started.

Then there is a step 202 executed.

In step 202 information about controlled variables y is received from different controlled systems. In the example, a vector y is determined that includes the controlled variables of the different controlled systems.

Then there is a step 204 executed.

In step 204 a trajectory of future states of the model up to a time horizon N _{p is} determined for a current state of the model and the at least one manipulated variable u is determined as a function of the time-discrete dynamic model of the vehicle to be controlled.

In the example, the model predictive controller is defined as a function of the change in the at least one manipulated variable Δu for each time step up to the time horizon N _p .

The model-predictive controller carries out a time-discrete state prediction which is defined by the linear time-invariant state model described and the Laguerre coefficients η _i .

The trajectory is preferably determined as a function of the quality measure J. In the example, the quality factor J depends on the control deviation e and the control effort. In the example, the quality measure J is defined as a function of at least one limit value which defines a limitation of an adjustment range of an actuator. The limit value can also characterize a limit for the value of one of the manipulated variables u.

Then there is a step 206 executed.

In step 206 the at least one manipulated variable u for controlling at least one of the controlled systems is specified. Different manipulated variables u influence different vehicle components in the example. Vehicle components are, for example, the electric drive in particular, the rear axle steering, a control unit and / or actuators for torque vectoring, an all-wheel drive system, and a chassis actuator.

Then the step 202 executed. The method ends, for example, when the vehicle is switched off.

The vehicle components are influenced in particular in an autonomous driving mode.

Claims (7)

Method for regulating a vehicle, characterized in that a model predictive controller specifies at least one manipulated variable (u) for at least one of the plurality of controlled systems for influencing the driving behavior of a vehicle, the at least one manipulated variable (u) depending on a time-discrete dynamic model of the vehicle to be controlled is determined (204), with a trajectory of future states of the model up to a time horizon being determined (204) for a current state of the model, the trajectory depending on a change in the at least one position variable (Δu (k)) is determined, and the change in the at least one manipulated variable (Δu (k)) for determining the trajectory is approximated by means of at least one Laguerre function, characterized in that the trajectory is dependent on a quality measure (J ) is determined, where the quality measure (J) depends on a control deviation (e) and a control effort, and that the quality aß (J) is defined as a function of at least one limit value, which characterizes a limitation of a control range of an actuator or a limitation for the value of a manipulated variable (u) for at least one of the plurality of controlled systems. Method according to one of the preceding claims, characterized in that the model-predictive controller comprises a time-discrete state prediction which is defined by a linear time-invariant state model and Laguerre coefficients (η _i ). Method according to one of the preceding claims, characterized in that the model predictive controller is defined as a function of the change in the at least one manipulated variable (Δu (k)) for each time step (k) up to a time horizon (Np). Method according to one of the preceding claims, characterized in that different manipulated variables (u) influence different vehicle components from the drive, rear-axle steering, control unit and / or actuators for torque vectoring, all-wheel drive system, chassis actuator, in particular in autonomous driving. Method according to one of the preceding claims, characterized in that the at least one manipulated variable (u) is determined as a function of a reference variable. Device (100) for controlling a vehicle, characterized in that the device (100) comprises a microprocessor (102) and a memory (104) with commands, when they are executed by the microprocessor (102) the method according to one of the Claims 1 to 5 is performed. Device (100) after Claim 6 , characterized in that the device (100) comprises a first interface (106) for at least one controlled variable (y) and a second interface (108) for at least one manipulated variable (u), the first interface (106) being designed as information to receive controlled variables (y) from different controlled systems, the second interface (108) being designed to output manipulated variables (u) for different controlled systems.

Images